Predicting simulator-based driver training impact on fuel efficiency

Although simulator-based training has been proven to help reduce fuel consumption in vehicles, the consistent nature of the associated improvement in fuel efficiency as well as the decay in the learning impact over time is yet to be fully defined. The aim of this paper is to predict the probable fuel efficiency improvements based on a simulator-based training program targeting a 1,200 vehicle taxi fleet operation. Numerical analysis of 450,000 records of fleet fueling data was performed in order to determine the fuel efficiency specifications of the fleet in order to enhance the specificity of the prediction framework. An analysis of the improvement and decay functions associated with the program demonstrated an expected improvement of 2.64% over the entire year following training. Statistical analysis relating to the population size and post-training monitoring period showed that in order to achieve 95% statistical certainty, a minimum population size of 20 drivers should be utilized with a minimum monitoring duration of 60 days. The next phase of the project will involve implementing the training program according to the criteria mentioned above in order to validate the results of this initial analytical phase.